acr2csm#
- class diffsptk.AutocorrelationToCompositeSinusoidalModelCoefficients(csm_order)[source]#
See this page for details.
- Parameters:
- csm_orderint >= 0
Order of CSM coefficients, \(M\).
References
[1]S. Sagayama et al., “Duality theory of composite sinusoidal modeling and linear prediction,” Proceedings of ICASSP, pp. 1261-1264, 1986.
- forward(r)[source]#
Convert autocorrelation to CSM coefficients.
- Parameters:
- rTensor [shape=(…, M+1)]
Autocorrelation.
- Returns:
- outTensor [shape=(…, M+1)]
Composite sinusoidal model coefficients.
Examples
>>> x = diffsptk.nrand(4) >>> x tensor([ 0.0165, -2.3693, 0.1375, -0.2262, 1.3307]) >>> acorr = diffsptk.Autocorrelation(5, 3) >>> acr2csm = diffsptk.AutocorrelationToCompositeSinusoidalModelCoefficients(3) >>> c = acr2csm(acorr(x)) >>> c tensor([0.9028, 2.5877, 3.8392, 3.6153])